Pd-Supported Catalysts: Evolution of Support Porous Texture along Pd
Deposition and Alkali-Metal Doping
R. Pellegrini,*
,†G. Leofanti,
†,§G. Agostini,
‡E. Groppo,
‡M. Rivallan,
‡,)and C. Lamberti
‡ †Chimet S.p.A - Catalyst Division, Via di Pescaiola 74, I-52041 Viciomaggio (Arezzo), Italy,‡Department of Inorganic, Physical and Materials Chemistry, NIS Centre of Excellence and INSTM Reference Center, University of Turin, Via P. Giuria 7, I-10125 Torino, Italy,§Consultant, Via Firenze 43, 20010 Canegrate (Milano), Italy, and)
Ecole Polytechnique Federale de Lausanne (EPFL), GGRC-ISIC, Station 6, CH-1015 Lausanne, Switzerland Received January 8, 2009. Revised Manuscript Received February 19, 2009
Adsorption of N2at 77 K and scanning electron microscopy have been used to measure the changes in the support
morphology, at nano- and microscale level, along the processes involved in the preparation of a supported Pd catalyst: Pd deposition, doping, and thermal treatments. Among the investigated supports, viz., activated carbons,γ-Al2O3,
SiO2, and SiO2-Al2O3(SA), the SA one was found particularly sensitive to these processes, as a result of its high
plasticity and reactivity. Involved processes can be summarized as follows: (i) During the Pd deposition, the support itself is partially dissolved and removed as a result of both the basicity of the precipitating agent and the final washing. (ii) When the undoped sample is thermally treated up to 823 K, only modest phenomena are observed. (iii) Upon doping with potassium carbonate, the support dissolution continues, and the greater the carbonate concentration, the greater the dissolution extent. In this case the dissolved material is not removed, but reprecipitates (partially outside the pores), during the subsequent drying at 393 K. (iv) When doped samples are thermally treated, the reaction between carbonate and support causes the mobilization of the support itself, with sintering phenomena that can reach the total collapse of the porous structure. The starting temperature of the pore collapse decreases with increasing potassium carbonate concentration. The modification of the support influences, directly or indirectly, the surface properties and the availability of Pd particles that can be doped or even covered by materials from support and made more or less accessible or even inaccessible by pore narrowing, widening, or blocking.
1. Introduction
Precious metal catalysts are widely used in the supported form on porous materials such as activated carbon,γ-alumina, silica, silica-alumina, organic polymers, and so forth, with a variety of properties (e.g., particle size, surface area, pore volume and size, mechanical strength and resistance to attrition, thermal stability, etc.). The final catalytic properties depend on the combination of the types of metal (i.e., the active phase)1-4 and support.2,4-7 Particularly, the support can exhibit a direct influence on the catalytic reaction, as its surface is often active toward reactants and reaction products, but an indirect influence also occurs in that the physical-chemical properties of the support influence the metal dispersion, its resistance to sintering, and the accessibility of active sites to reactants. Some of these aspects have been
inves-tigated in the literature, such as the effect of the support chemical properties (e.g., acidity or basicity)8-10or porosity.11,12
However, it is often neglected that, during the catalyst synth-esis, the original bare support may undergo modifications, at the nano- and microscale level, along with the entire catalyst pre-paration process that involves metal deposition, washing, thermal treatments, addition of additives, and so forth. For example, some carriers such as activated carbons, can release ashes when contacted with impregnating or washing solutions.5Some others, such as oxidic supports, are partially soluble in acidic and basic solutions used for the deposition of the metal,13-16 and the dissolved material can be washed out or can reprecipitate inside the pores. Several supports are sensitive to thermal treatments such as silica,17,18alumina,19titania,20zirconia,21and zeolites,22 or this sensitivity can be induced by the modifications to the chemistry of the support occurring in the previous preparation steps.
*Corresponding author. Phone: +390575-441565. Fax: +390575-441424. E-mail: riccardo.pellegrini@chimet.com.
(1) Wiersma, A.; van de Sandt, E. J. A. X.; den Hollander, M. A.; van Bekkum, H.; Makkee, M.; Moulijn, J. A. J. Catal. 1998, 177, 29–39.
(2) Panagiotopoulou, P.; Kondarides, D. I. Catal. Today 2006, 112, 49–52. (3) Hegedus, L.; Mathe, T.; Tungler, A. Appl. Catal., A 1996, 147, 407–414. (4) Bernas, A.; Kumar, N.; M::aki-Arvela, P.; Kul’kova, N. V.; Holmbom, B.; Salmi, T.; Murzin, D. Y. Appl. Catal., A 2003, 245, 257–275.
(5) van de Sandt, E. J. A. X.; Wiersma, A.; Makkee, M.; van Bekkum, H.; Moulijn, J. A. Appl. Catal., A 1998, 173, 161–173.
(6) Pinna, F.; Menegazzo, F.; Signoretto, M.; Canton, P.; Fagherazzi, G.; Pernicone, N. Appl. Catal., A 2001, 219, 195–200.
(7) Cizmeci, M.; Musavi, A.; Tekin, A.; Kayahan, M. J. Am. Oil Chem. Soc. 2006, 83, 1063–1068.
(8) Markus, H.; Plomp, A. J.; Maki-Arvela, P.; Bitter, J. H.; Murzin, D. Y. Catal. Lett. 2007, 113, 141–146.
(9) Venezia, A. M.; La Parola, V.; Pawelec, B.; Fierro, J. L. G. Appl. Catal., A 2004, 264, 43–51.
(10) Ruta, M.; Semagina, N.; Kiwi-Minsker, L. J. Phys. Chem. C 2008, 112, 13635–13641.
(11) Min, K.-I.; Choi, J.-S.; Chung, Y.-M.; Ahn, W.-S.; Ryoo, R.; Lim, P. K. Appl. Catal., A 2008, 337, 97–104.
(12) Okhlopkova, L. B.; Lisitsyn, A. S.; Likholobov, V. A.; Gurrath, M.; Boehm, H. P. Appl. Catal., A 2000, 204, 229–240.
(13) Brunelle, J. P. Pure Appl. Chem. 1978, 50, 1211–1229.
(14) Bernal, S.; Calvino, J. J.; Lopez-Cartes, C.; Pintado, J. M.; Perez-Omil, J. A.; Rodrıguez-Izquierdo, J. M.; Hayek, K.; Rupprechter, G. Catal. Today 1999, 52, 29–43.
(15) Balint, I.; Miyazaki, A.; Aika, K. Chem. Mater. 2001, 13, 932–938. (16) Shah, A. M.; Regalbuto, J. R. Langmuir 1994, 10, 500–504.
(17) Bossi, A.; Leofanti, G.; Moretti, E.; Giordano, N. J. Mater. Sci. 1973, 8, 1101–1109.
(18) Guo, S.-L.; Arai, M.; Nishiyama, Y. Appl. Catal. 1990, 65, 31–44. (19) Arai, H.; Machida, M. Appl. Catal., A 1996, 138, 161–176.
(20) Yu, J. G.; Yu, J. C.; Ho, W. K.; Leung, M. K. P.; Cheng, B.; Zhang, G. K.; Zhao, X. J. Appl. Catal., A 2003, 255, 309–320.
(21) Mercera, P. D. L.; Vanommen, J. G.; Doesburg, E. B. M.; Burggraaf, A. J.; Ross, J. R. H. Appl. Catal. 1990, 57, 127–148.
(22) Liu, S.-B.; Wu, J.-F.; Ma, L.-J.; Tsai, T.-C.; Wang, I. J. Catal. 1991, 132, 432–439.
In addition, it is a common practice to modify the catalytic properties by means of proper additives called promoters, modi-fiers, or dopants.23-25For example, some catalytic processes such as the hydrogenation of phenol to cyclohexanone,26-30 the selective hydrogenation of acetylene to ethylene,31 and the conversion of syn-gas to methanol32-34require alkali and alkaline earth-doped palladium catalysts. Even in this case, a support can undergo modifications through contact with the doping solu-tions.17,19,35It turns out that the modifications of the support during the catalyst preparation can affect the availability of the metal particles for the catalytic process to take place. In fact, the metal particles may be covered by dissolved support material, or by the shrinkage of the porous system of the support. Further-more, the metal particles can undergo doping by foreign species other than the added doping salt, but coming from the support.
During our studies on catalyst preparation and doping, we have frequently faced problems of support modification, for example, on activated carbons and oxides, as alumina, silica, and silica-alumina. In all the cases, the investigation of the changes undergone by the support upon different operations allowed us to better understand the behavior of the supported metal. In this paper we report the results of a systematic study on a sodium-neutralized silica-alumina support, particularly sensitive to these processes, showing how palladium deposition, thermal treatments, and doping with potassium salts lead to changes of the support morphology and, thus, of the catalyst properties.
The study has been carried out mainly by adsorption of N2at 77 K, and scanning electron microscopy (SEM), in order to assess any morphological changes at the nano- and microscale level, respectively. Also transmission electron microscopy (TEM), temperature-programmed desorption coupled with mass metry (TPD-MS), Fourier transform infrared (FTIR) spectro-scopy, and CO chemisorption have been used to better understand the phenomena occurring on the support and their influence on the properties of the metal particles. This last aspect will be presented in more detail in two subsequent papers.36,37
2. Experimental Details
2.1. Materials. Several supports were investigated: γ-alumi-na (A), sodium-neutralized silica-alumina (SA) with a SiO2/Al2O3ratio of 5.7, silica (S), and three types of activated
carbons from wood (CW), coconut shells (CC) and peat (CP). All the supports were in powder form with the exception of CC, which was in granules.
Pd precursor was deposited on the supports by the following different procedures:
(a) Pd catalysts supported on A and SA were prepared by the deposition-precipitation (D-P) method,38,39 using Na2PdCl4 as the palladium precursor and
Na2CO3as the basic agent. The supported palladium
hydroxide was then water-washed until residual chlorides were removed, and dried at 393 K over-night. The final Pd/A undoped catalyst contained 1.70% Pd and traces of Na. The Pd/SA undoped catalyst contained 2.05% Pd and some impurities arising from SA (Na 2.64%, K 0.04%, Ca 0.05%, Mg 0.04%, S < 0.10%). All percentages given above are referred to wt %; the same holds throughout the whole text.
(b) Pd/S catalyst was prepared by dry impregnation (DI) of microspheroidal silica with Pd(NH3)4(NO3)2
solu-tion followed by drying at 393 K and calcinasolu-tion at 723 K.32The final Pd/S catalyst contained 1.93% Pd and Na < 0.05%.
(c) Pd/CW and Pd/CP were prepared by the D-P method as described above, the Pd loading being 5.0%. (d) Pd/CC was prepared by wet impregnation (WI) with an
aqueous solution of Na2PdCl4 and further chemical
reduction as described in ref 40. The supported palla-dium phase was then water-washed until residual chlor-ides were removed, and dried at 393 K overnight. The Pd loading was 0.5%.
In the case of Pd/oxide systems, the Pd deposition was followed by K-doping. K-doped Pd/A and Pd/S catalysts were obtained by DI of corresponding undoped catalysts with an aqueous solution of potassium carbonate, followed by drying in a static oven for 16 h at 393 K. The potassium concentration on the final doped catalysts was 3%. The Pd/SA catalyst was doped with K by the same method using different dopant compounds, i.e., carbonates and chlorides, in order to obtain 3-12% of potassium concentration in the final catalyst.
A further thermal treatment at 823 K was performed on all oxide-based catalysts, both undoped and doped, in a static oven for 30 min. Pd/SA was also treated at other temperatures in the 473-1073 K range with the same procedure, and at 673 and 823 K in flowing H2in a quartz reactor. Finally, some doped
catalysts were submitted to washing in order to remove the soluble fraction. Washing was performed with water at 318 K inside a sintered disk filter funnel.
For the 2% Pd/SA systems, a proper nomenclature has been adopted to directly identify the approximate chemical composi-tion of the sample. Thus, undoped 2% Pd/SA is labeled as Pdund.
For samples doped with potassium carbonate, we add the name of the dopant cation followed by the atomic ratio K/Pd. So, for example, PdK4 represents a sample containing 2% Pd and potassium from carbonate salt at K/Pd = 4 at/at. If alkali metal dopants are introduced as compounds other than carbonate, such as chloride, the counterion is indicated by the suffix (Cl), for example PdK(Cl)4.
2.2. Characterization Techniques. 2.2.1. Adsorption of N2at 77 K.The surface area, pore volume and pore size
distribution of both catalysts and supports were measured by adsorption of nitrogen at 77 K on samples pretreated at 393 K under vacuum for 2 h. Measurements were performed on a
(23) Ponec, V. Appl. Catal., A 1997, 149, 27–48. (24) Murata, S.; Aika, K.-I. J. Catal. 1992, 136, 110–117.
(25) Pinxt, H. H. C. M.; Kuster, B. F. M.; Marin, G. B. Appl. Catal., A 2000, 191, 45–54.
(26) Neri, G.; Visco, A. M.; Donato, A.; Milone, C.; Malentacchi, M.; Gubitosa, G. Appl. Catal., A 1994, 110, 49–59.
(27) Scire, S.; Crisafulli, C.; Maggiore, R.; Minico, S.; Galvagno, S. Appl. Surf. Sci. 1996, 93, 309–316.
(28) Scire, S.; Minico, S.; Crisafulli, C. Appl. Catal., A 2002, 235, 21–31. (29) Mahata, N.; Raghavan, K. V.; Vishwanathan, V. Appl. Catal., A 1999, 182, 183–187.
(30) Pillai, U. R.; Sahle-Demessie, E. Appl. Catal., A 2005, 281, 31–38. (31) Park, Y. H.; Price, G. L. Ind. Eng. Chem. Res. 1992, 31, 469–474. (32) Gusovius, A. F.; Watling, T. C.; Prins, R. Appl. Catal., A 1999, 188, 187–199. (33) Kazi, A. M.; Chen, B.; Goodwin, J. G.; Marcelin, G.; Rodriguez, N.; Baker, T. K. J. Catal. 1995, 157, 1–13.
(34) Gusovius, A. F.; Prins, R. J. Catal. 2002, 211, 273–277.
(35) Mercera, P. D. L.; Vanommen, J. G.; Doesburg, E. B. M.; Burggraaf, A. J.; Ross, J. R. H. Appl. Catal. 1991, 71, 363–391.
(36) Agostini, G.; Pellegrini, R.; Leofanti, G.; Bertinetti, L.; Bertarione, S.; Groppo, E.; Zecchina, A.; Lamberti, C. J. Phys. Chem. C, submitted for pub-lication, 2009.
(37) Pellegrini, R.; Leofanti, G.; Agostini, G.; Bertinetti, L.; Bertarione, S.; Groppo, E.; Zecchina, A.; Lamberti, C. J. Catal., submitted for publication, 2009.
(38) Geus, J. W.; van Dillen, A. J. Preparation of supported catalysts by deposition-precipitation. In Handbook of Heterogeneous Catalysis; Ertl, G., Kn::ozinger, H., Weitkamp, J., Eds.; Wiley-VCH: Weinheim, Germany, 1997; Vol. 1, pp 240-257.
(39) Simonov, P. A.; Likholobov, V. A. Physicochemical aspects of preparation of carbon-supported noble metal catalysts. In Catalysis and Electrocatalysis at Nanoparticle Surfaces; Wieckowski, A., Savinova, E. R., Vayenas, C. G., Eds.; Marcel Dekker, Inc.: New York, 2003; pp 409-454.
(40) Menegazzo, F.; Fantinel, T.; Signoretto, M.; Pinna, F. Catal. Commun. 2007, 8, 876–879.
Micromeritics ASAP 2000 instrument. Once the adsorption isotherm was obtained, the following calculation methods were used:41,42
1. Brunauer-Emmett-Teller (BET) equation in the 0.06-0.20 p/p0interval for the determination of surface area (As), and of the parameter c that depends on the
inter-action between the first monolayer of adsorbed N2and
the solid surface;
2. Gurvitsch method, i.e., the measurement of the N2
volume adsorbed at p/p0= 0.973 by using a conversion factor for volume of liquid N2into pores = 1.546 (mm3
liquid/cm3STP vapor), for the determination of the total
pore volume (Vp);
3. Equation Dm = 4(Vp/As) for the determination of
mean pore diameter (assumed as cylindrical) for oxide supports and oxide-based catalysts, and equation Dm= 2(Vp/As) for activated carbons and corresponding
catalysts, assumed to have slit-shaped pores;
4. Barrett-Joyner-Halenda (BJH) method with the Harkins-Jura equation for the calculation of the thick-ness of the N2adsorbed layer, used for the determination
of pore distribution of oxide and oxide-based catalysts, here also assumed to have cylindrical pores.
All the data were normalized to the weight of Pd catalysts dried at 393 K in static air, in order to take into account the effect of matter increase due to Pd and dopant addition, and matter loss caused by thermal treatments.43The weight changes caused by thermal treatments were measured at various temperatures (393-1073 K) in different atmospheres (air and H2), depending
on the catalyst. Before each treatment, all the samples were dried at 393 K in a static oven for 16 h. The treatments in air consist of heating the sample inside a quartz crucible in a static oven up to the treatment temperature at 10 K min-1and keeping it at this temperature for 30 min. The treatments in H2consist of heating
the sample inside a quartz tube up to the treatment temperature at 10 K min-1in He, switching to H2, and keeping the sample at
this temperature for 30 min. In both cases, the sample was cooled to room temperature (RT) and weighted.
2.2.2. Scanning Electron Microscopy.SEM observation was carried out by a Philips XL20 instrument at 30 kV of voltage. The dry sample was put on an aluminum stub covered with a double-layer adhesive disk and then sputtered with gold for 120 s. Electron dispersive spectroscopy (EDS) analysis was performed by an annexed EDAX operating at 15 kV.
2.2.3. Transmission Electron Microscopy. High-resolution transmission electron micrographs were obtained with a JEOL 3010-UHR operating at 300 kV, equipped with a 2k 2k pixels Gatan US1000 CCD camera. Samples were deposited on a copper grid covered with a lacey carbon film.
2.2.4. FTIR Spectroscopy.The FTIR spectra were recorded at 300 K at 2 cm-1resolution, using a Bruker IFS 28 spectro-meter, equipped with a cryogenic mercury-cadmium-telluride (MCT) detector. Samples were analyzed in the form of self-supported pellets inside an IR quartz cell allowing in situ thermal treatments and gas dosages.
2.2.5. Temperature Programmed Desorption Coupled with Mass Spectrometry.TPD was monitored by mass spectrometry on samples previously activated in inert (He) atmosphere at 373 K for 4 h (in order to remove most of physisorbed water). During the increase of temperature from RT to 873 K at a ramp of 5 K min-1
under inert flow (1.0 vol % Ar (as tracer) in He at a constant flow rate of 50 mL min-1), the following amu (m/e) were collected: 18 (H2O), 28 (CO/N2), 32 (O2), 40 (Ar), and 44 (CO2).
2.2.6. CO Chemisorption.CO chemisorption experiments were performed by the dynamic pulse method at 323 K. The catalyst, previously dried at 383 K overnight, was reduced in situ at 393 K with the following procedure: (i) the sample was loaded inside a U-tube, (ii) it was heated in He up to 393 K (with a 10 K min-1ramp rate), (iii) H2was fed for 30 min, and (iv) the sample was finally
cooled to 323 K in He (with a 10 K min-1ramp rate).
After that, five CO pulses of known volume were sent to the sample under He flow, and the adsorbed CO amount was determined by a TCD detector. In order to determine the available surface Pd atoms (surface to volume ratio, S/VChemi)
a CO/Pd average stoichiometry of 1 was assumed. This assump-tion was verified by a parallel series of measurements performed on three different samples treated at two different temperatures with H2-O2 static volumetric titration, which gives an O/Pd
average stoichiometry close to 1.44The two methods/techniques give a CO/O ratio in the 0.94-1.13 range, which is a strong support for the assumption that CO/Pd average stoichiometry equals 1. A further confirmation of the assumed stoichiometry comes form the Pdundsample reduced in H2at 393 K. In this case,
the fraction of surface atoms derived from the Pd particle size distribution measured by TEM is 0.29, a value very close to the value of 0.28 obtained from CO chemisorption assuming a CO/Pd average stoichiometry of 1.
3. Results and Discussion
3.1. Support Modifications upon Catalyst Preparation and Treatments. Table 1 reports the results of nitrogen adsorp-tion measurements on catalysts prepared with different methods and on different supports. Measurements have been carried out on undoped as well as on doped samples either dried at 393 K or calcined at 823 K.
Compared to all the oxidic supports, the activated carbons are more inert toward Pd deposition with different methods (i.e., D-P or WI): only small variations of Asand Vpare observed
Table 1. Effect of Preparation (i.e., Palladium Deposition by D-P, DI, and WI, and Subsequent Doping with Potassium) and Thermal
Treatments on Surface Area (As), Pore Volume (Vp) and Mean Pore
Size (Dm) of Catalysts Supported on Different Carriersa
as such Pd impregnated K doped (3% K) parameter support 393 K method 393 K 823 K 393 K 823 K As(m2g-1) A 177 D-P 200 186 190 176 S 320 DI 314b 310 151 69 SA 179 D-P 141 124 99 76 CW 982 D-P 981 CC 1455 WI 1459 CP 876 D-P 908 Vp(mm3g-1) A 467 D-P 437 463 407 416 S 1138 DI 1127b 1104 1080 999 SA 719 D-P 827 716 839 655 CW 551 D-P 521 CC 510 WI 500 CP 672 D-P 684 Dm(A˚) A 106 D-P 87 100 86 94 S 142 DI 144b 143 286 579 SA 160 D-P 235 231 341 346 CW 11 D-P 11 CC 7 WI 7 CP 15 D-P 15 a
All thermal treatments (at 393 or 823 K) have been carried out in air. The data have been normalized to the weight of undoped catalyst dried in air at 393 K.bTreatment temperature 723 K (see Section 2. Experi-mental Details).
(41) Gregg, S. J.; Sing, K. S. W. Adsorption, Surface Area and Porosity, 2nd ed.; Academic Press, Inc.: New York, 1982.
(42) Rouquerol, F.; Rouquerol, J.; Sing, K. Adsorption by Powders & Porous Solids. Principles, Methodology and Applications; Academic Press: New York, 1999.
(43) As an example, the weight loss for calcination at 823 K of the Pdundsample
isΔWt = [Wt(393) - Wt(823)]/Wt(823) = 0.0515. This means that, to estimate the original weight of a sample calcined at 823 K, i.e., Wt(873)norm, the weight
measured after calcination has to be multiplied by a factor 1.0534: Wt(873)norm
upon passing from carbon to supported catalyst. Conversely, all the oxidic supports are more sensitive than carbons toward the Pd addition and exhibit relevant variations upon thermal treat-ment and/or doping. The extent of the variation depends on the oxide type.
Specifically, as a result of Pd deposition on Al2O3support, As increases, and Vpdecreases with consequent reduction of Dm, confirmed by the pore size distribution (not reported for brevity). Part of the effect is due to the rehydration of alumina. However, a role of etching by basic medium used in the D-P method is highly probable, even if the mechanism is unknown. Upon calcination, the sample shows a loss of Aswith contem-porary increase of Dm, indicating that the solid has undergone the early effects of sintering process, characterized by growth of larger elementary particles (i.e., γ-alumina crystals) at the expense of the smaller ones, so that their total area is smaller and the intraparticles spaces (i.e., the pores) are larger.19The doping procedure seems to have a smaller effect than Pd deposition, bringing about a moderate loss of Asand Vp. Since the carrier was contacted with a basic solution in both cases, (i.e., Pd deposition and doping), the different behavior can be explained in two different ways: (i) during impregnation the surface irregularities, more sensitive to chemical attack, are quite completely removed, making the second exposure to basic solution during doping less effective, or (ii) the effect is more evident upon palladium deposition because the subsequent washing removes the dissolved material that, conversely, repre-cipitates upon drying after doping. Thermal treatment of doped sample has an effect similar to the same treatment on the undoped one: the mechanism is the same and is only slightly influenced by the presence of the dopant, at least up to 823 K.
The SiO2 support is not significantly modified during the impregnation with Pd because the preparation method (i.e., DI with Pd(NH3)4(NO3)2solution) avoids the contact of the sup-port with large volumes of aggressive solutions. Undoubtedly, if the D-P method had been used instead of the DI one, the low resistance of SiO2toward alkaline solution would have caused heavy modifications of its porous structure. In fact, this occurs
upon doping, as this is carried out by filling the pores with K2CO3solution followed by drying at 393 K: a collapse of Asat almost constant Vpand an enlargement of pores is observed. This can be explained either by the formation of soluble silicates that enable the destruction of the pore walls, or by the solution-precipitation mechanism, which enables the growth of the silica elementary particles. The effect of thermal treatments strongly depends on the presence of dopant: while the undoped catalyst is almost stable at 823 K, the doped one undergoes a sintering with enlargement of the pores and, consequently, a sharp decrease of As.
The behavior of SA support is interesting in that it undergoes consistent variations of the morphology at the nano- and microscale level, and this is the reason why we have decided to deeply investigate such support. In fact, it is observed that the use of basic solutions for Pd deposition and doping brings about an enhancement of pore volume and diameter accompanied by a decrease in As. Moreover, a decrease in Asand Vpat a constant pore diameter is observed upon thermal treatment at 823 K. This peculiar behavior is discussed in the following sections.
3.2. Processes Occurring during Palladium Deposition on SA. Table 1 indicates that the SA support used for the preparation of the catalysts under investigation has a low resistance toward the solution used during palladium deposi-tion: Asdecreases, while Vpand Dmincrease; the last parameter is in agreement with the pore size distribution, which undergoes a general enlargement (Figure 1a). The process that causes such a phenomenon can be explained only with the removal of matter from the inner pores, as will be shown in section 3.4. However, this cannot be thought of as a simple enlargement of pores due to the solubilization of a layer of their walls, because, in that case, we should observe an increase in Vpgreater than the increase in Dm, and also an increase of As. For example, on pores with cylindrical shape, the volume is directly proportional to Dm2, while the surface is directly proportional to Dm. It seems more likely that the solubilization involves the removal of irregula-rities or diaphragms inside the pores (see, for example, the model of Figure 10a): to a small increase in volume (the volume of
Figure 1. Pore size distribution of undoped, doped, and washed catalysts treated at different temperatures in air or H2: (a) SA and Pdund
samples; (b) effect of doping at 393 K; (c) effect of thermal treatments on PdK4 sample; (d) effect of thermal treatments and washing on PdK20
sample. In panel a, the curves relative to the treatments in air at 473, 573, 673, and 923 K, and in H2at 393 and 673 K, very similar to those
reported, have been omitted for the sake of clarity. Some curves are reported in more than one part in order to make more evident the shifts of pore size distribution.
removed matter) corresponds a strong decrease of area (the surface of irregularities or diaphragms) and a real, or apparent, increase in diameter.
Notice that the process hypothesized to occur on SA is totally different from what we have observed on alumina (Table 1), where the support solubility is certainly lower: Asincreases and Vpdecreases with consequent decrease of Dm.
3.3. Processes Occurring during Calcination and Reduc-tion of Pdund Sample. During calcination, a weight change occurs in the Pdundsample, and thus, when different samples are compared, some apparent variations of Asand Vpare actually due to the “concentration” of the sample caused by loss of water. The problem has been solved by normalizing the data at a given state of the sample, taking into account its composition at the initial stage and after subsequent changes. For this purpose, the sample Pdund dried in air at 393 K has been selected as a reference (see Section 2. Experimental Details).
The effect of calcination temperature on the Pdundsample has been deeply investigated by systematically measuring As, Vp, and Dm in the 393-1073 K range, as reported in Table 2. Calcination up to 923 K causes a moderate decrease of Asand Vp, while the pore size (Dm) remains almost unchanged, as confirmed by the pore size distribution (Figure 1a). The most significant changes occur between 393 and 573 K, suggesting that this is not a phenomenon of classical sintering, which would require higher temperature. On the basis of the results obtained by weight loss measurements and TPD-MS, we can associate the process with the loss of water that is maximum in this range of temperature (Figure 2a) and far more relevant than the release of other substances, such as CO2. Figure 2b shows that there is a good correlation between the two processes, i.e., surface area decrease and water loss. Notice that between 673 and 923 K, where most of water has been already released and the loss of weight is low, the variation of Asis negligible. Moreover, the constancy of Asconfirms that sintering is not operating at lower temperatures, but it requires heating at temperatures higher than 923 K (Table 2). A further confirmation of this picture comes from the monotonous decrease in the value of the c parameter (from 151 at 393 K to 123 at 573 K) as the temperature of thermal treatment increases, quite similar to what is observed for alumina (from 134 at 393 K to 97 at 823 K), where the only phenomenon that takes place in this range of temperature is the dehydration with loss of hydroxyl groups from the surface. The analogy of the variation of the c parameter for the two types of supports suggests that, in the Pdund sample treated at high temperatures, no real chemical reactions between components or decompositions with relevant release of matter are taking place, as these are impossible on alumina. The collapse in As observed on a sample treated at 1073 K is attributable to the presence of residual alkali metal and will be discussed in detail in section 3.5.
In agreement with this picture, significant differences between treatments in air and H2 are not observed (Table 2 and Figure 1a), pointing out that all the differences induced by the different treatment atmosphere on Pd crystal characteristics and catalytic behavior36,37 are not attributable to a different nanoscale morphology.
The observed variations between 393 and 923 K are too small to define a model based on changes in the local morphology of a solid at the pore level. At most, we can assume that, as a result of dehydration and high temperature treatments, one or both of the following phenomena are apt to occur: (i) occlusion of narrow connections between pores, which leads to a decrease of Vpand As, but does not change Dm, under the hypothesis that the pore blocking is independent of its size; in this case, a significant movement of matter is not involved; (ii) slight shrinking (implying a parallel decrease of Vpand Dmwith lower decrease of As) counterbalanced by a smoothing of the pore’s inner surface (implying a decrease of Asat constant Vp, with a consequent increase of Dm); in this case a consistent movement of matter is involved. In Figure 10b a possible model that brings together both mechanisms is proposed.
Finally, notice that the evolution of As, Vp and Dm with temperature observed on alumina (Table 1) is quite different: in this case, a mechanism of moderate sintering takes place with growth of elementary larger particles (i.e.,γ-alumina crystals) at the expense of the smaller ones, so that their total area is smaller and the intraparticle spaces (i.e., the pores) are larger.
3.4. Processes Occurring during Doping of Pdund Sam-ple. As mentioned above, in order to properly analyze the data of doped catalysts, it should be noted that the addition of dopant causes a “dilution” of the catalyst by the added salt, which is certainly nonporous, and thus causes an apparent decrease of As and Vpper unit weight of solid. Thus, N2adsorption data have been normalized to the unit weight of catalyst Pdunddried in air at 393 K (Table 3).
Let us first consider the changes in the pore volume. If all the added matter (i.e., the carbonate) would be located inside the
Table 2. Adsorption of N2at 77 K on PdundCatalyst Submitted to
Different Thermal Treatmentsa
treatment air H2 temperature (K) 393 473 573 673 823 923 1073 393 673 823 As(m2g-1) 141 133 127 125 124 119 1 143 125 125 Vp(mm 3 g-1) 827 743 702 719 716 690 5 844 728 721 Dm(A˚) 235 223 222 230 231 232 237 233 231 c 151 147 123 127 110 109 153 131 116
aThe changes of surface area (A
s), pore volume (Vp), mean pore
diameter (Dm), and parameter c of the BET equation as a function of
atmosphere (air or H2) and temperature of thermal treatment are
reported.
Figure 2. (a) Weight loss at increasing heating temperature in air in
a static oven on the Pdundsample (gray curve) and H2O-TPD profile
of the same sample (black curve). (b) Surface area decrease on Pdund
pores, their volume would suffer a decrease at least equal to the volume of salt.45 Conversely, a slight upward trend of Vpis observed. Formally, this would indicate that the carbonate is outside the pores. However, this simple hypothesis presupposes the constancy of Asand Dmthat is not observed, as the former decreases and the latter increases. This means that the expansion of pores is real, as is confirmed by the distribution curves (Figure 1b).46
The trend is very similar to that previously observed going from support to catalyst: decrease of As, slight increase of Vp, and increase of Dmand c. In that case, it has been assumed that the dissolution concerns the irregularities or diaphragms inside the pores. Here, we can suppose that this process continues during the doping, probably involving adjacent pore walls. A possible objection is that the contact between the catalyst and a basic solution has already occurred during the palladium deposition step (where Na2CO3 is used) and so, the support irregularities should have been already removed and the effects of the new exposure should be less evident. However, the concentration of dopant K2CO3solution, especially on samples with higher K loading, is much greater than the Na2CO3 solution used for palladium deposition. Moreover, after Pd deposition, the catalyst is washed, and so, during the drying at 393 K, the pores are filled only by water. Conversely, doping is not followed by washing, and thus, during the drying step, the pore walls are in contact with a strong basic solution up to 393 K. The conclusion is that, during the catalyst doping, the support is subjected to a much more severe and aggressive environment than during the palladium deposition, and there-fore it is reasonable to expect a further chemical attack.
A confirmation that the observed behavior is due to the support dissolution by K2CO3has been obtained by two further measurements. The first concerns the catalyst doped with KCl instead of K2CO3, i.e., the PdK(Cl)4 sample: the solution of KCl has a neutral pH, so its attacking power toward SA support is rather low. Taking into account the weight correction for the KCl content, the PdK(Cl)4 sample exhibits values of As, Vp, Dm, and pore size distribution quite similar to the undoped sample: in particular, as shown in Table 3, the PdK(Cl)4 sample shows
the same Asand Dm(142 m2g-1, 242 A˚) as the undoped Pdund sample (141 m2g-1, 235 A˚) and is different from the K2CO3 -doped PdK4 sample (99 m2g-1, 341 A˚). Therefore, if the doping solution does not cause any support dissolution, the porous structure is not affected.
The second measurement concerns the catalyst PdK20(H2O), i.e., the PdK20 sample washed to remove the carbonate or other soluble compounds formed by reaction between carbonate and support (vide infra), and then dried at 393 K. If the changes observed on the catalyst before washing (i.e., PdK20) were due to a precipitation inside the pores of the soluble material from the dopant solution, the washing treatment should restore the initial situation. Indeed, the washing treatment leaves the alterations induced by the doping almost unchanged (Table 3): PdK20 and PdK20(H2O) have the same Asvalue (90 m2g-1) and Dm values that are close to each other (381 and 372 A˚, respectively). Consequently, the change at nanoscale morphol-ogy of the support upon doping is attributable to deep and irreversible changes in the support structure.47
In conclusion, a part of the support dissolves during doping, and then reprecipitates, at least partially, outside the pores during the subsequent drying, with consequent alteration of the whole porous structure. It is likely that the reprecipitated material is made, besides unreacted carbonate, by the product of reaction between carbonate and support (e.g., potassium sili-cate).17The reaction leads to the decomposition of carbonate with release of CO2and increase of sample weight lower than expected upon doping, suggesting that, already at the end of drying at 393 K, a fraction of carbonate is decomposed.48The nonreacted carbonate reprecipitates during the drying step, although, probably after the dissolved support because of its higher solubility.
The question is now where the reprecipitated material is located at this stage. Since the pore volume tends to increase slightly with increasing concentration of dopant (Table 3), the amount of material that entered the pores during doping is lower than the material expelled from the pores during drying. More-over, the material reprecipitated inside the pores is insoluble or slightly soluble in water, as confirmed by the fact that the washing of the PdK20 sample does not cause any increase in Vp, As, Dm, and pore size distribution. Consequently, we can suppose that the material reprecipitated inside the pores is made by the reaction product of support with carbonate, while the unreacted carbonate is located outside the pores. The phenom-enon is confirmed by the SEM observation (Figure 3) showing that the outer surface of the PdK4 granules is covered by
Table 3. Adsorption of N2at 77 K on Doped Catalysts Dried at 393 K in Aira
Sample
Pdund PdK4 PdK10 PdK20 PdK(Cl)4 PdK20(H2O)
treatment air 393 K air 393 K air 393 K + H2O + air 393 K
As(m2g-1) 141 99 104 90 142 90
Vp(mm3g-1) 827 839 851 854 858 833
Dm(A˚) 235 341 326 381 242 372
c 151 152 163 199 158 180
aThe changes of surface area (A
s), pore volume (Vp), mean pore diameter (Dm), and parameter c of the BET equation as a function of K content are
reported. The potassium doping salt is K2CO3apart from the PdK(Cl)4 sample doped with KCl. The PdK20(H2O) sample is the PdK20 submitted to
washing with water and further drying at 393 K.
(45) Considering the simplest model of pores, i.e., cylindrical blind pores, if carbonate is spread on the walls of the pores, their diameter decreases, causing a decrease of volume equal to the volume of salt and proportional to decrease of the diameter. If the carbonate occupies the bottom of the blind pores, the pore diameter does not change, and the volume decrease is still equal to the volume of salt. If the “cap” is over the bottom of the pores, the diameter does not change, and the closer the “cap” is to the pore mouth, the bigger the volume decrease.
(46) It should be noted that, while in most cases Dmand the maximum of pore
size distribution have similar values, samples with maximum distribution around 500-550 A˚ exhibit a Dmaround 380 A˚. The reason of this discrepancy is the
presence of a very small fraction of micropores, which contribute much more to As
than to Vp. There is neither information available in order to understand the origin
of such micropores (probably localized etching), nor indications that their presence nullifies the reported findings, so, in the following, this fact will no longer be discussed.
(47) The unchanged porous texture characteristics do not necessarily imply that the chemistry of the surface remains unchanged: the value of parameter c for the PdK20(H2O) sample (180) lies between the values of the unwashed PdK20 sample
(199) and the samples with lower K content (152-163).
(48) The decomposed fraction is roughly 80% in the PdK4 sample and 30% in the PdK20 one.
elongated crystals other than the support that are not observable before doping, constituted mainly by alkaline metal carbonates, as found by EDS analysis.
As far as the chemical nature of the surface is concerned, we observe that the value of c increases with increasing K/Pd ratio (Table 3). This is an effect only partially attributable to the contact between N2and the area covered by carbonate: washing with water brings about a decrease in the value of c, but not enough to return to the values of the undoped catalyst. The increase of c as a result of contact with carbonate and subse-quent washing is quite similar to that seen by passing from support to the Pd-supported catalyst (Table 1). In that case, the phenomenon has been attributed to a change in the chemistry of the support. In this case, we can invoke further modification of the chemistry of the Pdundsurface. In Figure 10c, a possible model is reported for the proposed mechanisms occurring during K doping.
3.5. Processes Occurring during Calcination and Reduc-tion of Doped Samples. We recall that thermal treatments have little effect on undoped catalyst (Table 2 and Figure 1a): a slight decrease of Asand Vpat constant diameter is observed, which has been interpreted as an occlusion of some narrow channels between contiguous pores, and/or a weak shrinking with smoothing of pores. The effect of thermal treatments on two doped catalysts (PdK4 and PdK20) are reported in Table 4 and shown in Figure 1c,d. Notice that, also in this case, the data have been corrected for both the carbonate content and the loss of H2O and CO2. For the sake of clarity, Figure 4 summarizes the evolution of Asand Vpas a function of temperature for different catalysts.
The effect of thermal treatments up to 1073 K can be divided into two zones. In the lower temperature range, a slight decrease of Asand Vpis observed, while Dm(not reported in Figure 4) is constant, as confirmed by the pore size distribution (Figure 1a,c, d). The range of relative stability of the samples decreases as K content increases: up to 923 K for Pdund, 823 K for PdK4 and 673 K for PdK20. At these temperatures, a sudden collapse of Vp
and Asoccurs. The decrease of Vpand Asis accompanied by a shrinking of pores (see, for example, sample PdK20 at 823 K in air) and, in much more stressed conditions (sample PdK4 at 1073 K in air) leads to the total disappearance of porous structure. The temperature at which the collapse occurs is inversely proportional to the K/Pd ratio: PdK20 < PdK4 < Pdund samples.
The observed changes are clearly due to the presence of carbonate, as confirmed by thermal treatments carried out on the sample PdK20(H2O), i.e., on the doped sample washed with water. In this case, the characteristics of the sample treated at 823 K in air are identical to those of the same sample simply dried at 393 K (see Figure 4): once the carbonate has been removed, the sample is no more affected by the thermal treatments.
The role of carbonate can be explained considering that carbonate becomes reactive with the support at high tempera-ture. In fact, it is well-known that the K carbonate (or Na carbonate) is used to dramatically lower the melting temperature of silica (the main component of the SA support) in the preparation of glasses. For example, sodium carbonate starts to react with silica already at 823 K,49 giving the following reaction: Na2CO3 + SiO2 f Na2SiO3 + CO2. The new, chemically modified support becomes mobile already at low temperature because it is well above the Tamman temperature of the pure support. At this point, its porous structure tends to collapse as a result of phenomena such as sintering, which is a process involving the elementary particles of a solid, which progressively assume a form that allows them to adapt to one another in order to gradually occupy the interparticle spaces (pores).50The silicate formed could react with the support not yet modified, thus exalting the effect of the process.
The destabilization of support through reaction with alkali metal carbonate requires (i) the disappearance of potassium
Figure 3. SEM images of the surface of catalyst particles: (a)
undoped catalyst Pdund; (b) doped catalyst PdK4.
Figure 4. Evolution of As(a) and Vp(b) as a function of thermal
treatment in air or H2for Pdund, PdK4, PdK20, and PdK20(H2O).
The majority of the data are from Tables 1-4.
(49) B::uchel, K. H.; Moretto, H.-H.; Woditsch, P. Silicate products. In Industrial Inorganic Chemistry; Wiley-VCH: Weinheim, Germany, 2000; pp 325-350.
(50) Randall, M. G. Sintering Theory and Practice; Wiley-Interscience: New York, 1996.
carbonate, which is otherwise stable at higher temperature,51 and (ii) the release of CO2. In the samples under investigation, the increase of thermal treatment temperature really causes the
progressive disappearance of carbonate (not yet decomposed after the drying at 373 K), as demonstrated by infrared spectro-scopy on PdK4 samples treated in the 373-823 K range (see Figure 5a). The sample treated at 373 K shows a band with a maximum at about 1430 cm-1, typical of the νas mode of carbonates.52 Treatments at higher temperatures cause a de-crease of band intensity (673 K) until the complete disappear-ance at 823 K. The process is accompanied by the release of CO2, as foreseen by reaction between carbonate and the support, and evidenced by TPD-MS experiment (Figure 5b).
The effect of “alkaline fusion” in the presence of potassium carbonate is confirmed by SEM observations that show a collapse of catalyst particles. In fact, the particles of the as-such PdK20, exhibiting a spherical shape, undergo a reduction of the size and a loss of the spherical shape when treated at 823 K (Figure 6).
A possible image of the processes described above is shown in the Figure 10d,e where the growth of support particles and the deep interaction between support and carbonate are evidenced: the interacted part, in light green, is a consistent part of the support, and even the entire support. Figure 10d represents the intermediate stage, where a collapse of Asis observed, but a part of the porous structure has resisted, as observed, for example, on the PdK20 sample treated at 823 K. In Figure 10e, the subse-quent total collapse of particles is represented, as occurred, for example, in the PdK4 sample calcined at 1073 K.
Table 4. Adsorption of N2at 77 K on Doped Catalysts As Such (PdK4 and PdK20) and Water Washed (PdK20(H2O)) Submitted to Different
Thermal Treatmentsa
sample
PdK4 PdK20 PdK20(H2O)
treatment air 393 K air 823 K air 1073 K air 393 K air 673 K air 823 K H2823 K air 393 K air 823 K
As(m2g-1) 99 76 0 90 75 46 65 90 90
Vp(mm3g-1) 839 655 0 854 743 352 484 833 840
Dm(A˚) 341 346 381 394 304 298 372 375
c 152 117 199 168 159 171 180 249
aThe changes of surface area (A
s), pore volume (Vp), mean pore diameter (Dm), and parameter c of the BET equation are reported.
Figure 5. (a) FTIR spectra of PdK4 in the region of carbonates as
a function of reduction temperature. (b) CO2-TPD profile of the
PdK4 sample, after previous pretreatment in He at 373 K for 4 h.
Figure 6. SEM images of the particles of doped PdK20 catalyst.
The as-such sample, i.e., treated only at 393 K (a), and 3
enlarge-ment (b); sample treated at 823 K (c) and 3 enlargement (part d).
Figure 7. Relationship between Pd dispersion and surface area on
Pd/SA system submitted to doping with K2CO3and to thermal
treatments in the 393-823 K temperature range. The curve
qua-litatively shows the dispersion versus Astrend for the samples
treated in air. Large and small symbols are associated with
treat-ments in air or H2atmosphere, respectively. Pdund, PdK4, PdK10,
and PdK20 data are reported as squares, circles, diamonds, and triangles, respectively; blue, green, and red colors refer to thermal treatments at 393, 673, and 823 K, respectively.
(51) Lehman, R. L.; Gentry, J. S.; Glumac, N. G. Thermochim. Acta 1998, 316,
Regarding the atmosphere of the thermal treatment, the H2 seems to have a slight effect on the stability of the PdK20 sample, in the sense that it seems to prevent the collapse of the porous structure. However, the difference between samples treated in air and H2at 823 K is moderate, so, being in a phase of great transformation, it can easily be justified by the different experi-mental setups used in the two cases.
3.6. Effect of Support Changes on Catalyst Properties. The change of the porous structure of the support during Pd deposition, doping, and thermal treatments affects all the characteristics of dispersed Pd crystals. This topic will be deeply discussed in two subsequent papers.36,37Here, some preliminary results are reported to evidence the direct relationship between support modification and Pd properties.
Figure 7 shows the variation of Pd dispersion (as obtained by CO chemisorption) versus As(as obtained by adsorption of N2 at 77 K) on different samples. Starting with the samples activated in air (larger symbols in Figure 7), it is evident that the decrease of Asis accompanied by a decrease in Pd dispersion, rather independently of the K/Pd atomic ratio and of the temp-erature necessary to obtain a given loss of surface area. The most relevant phenomenon that links the two processes is the cap-ability of the support to cover progressively palladium crystals, both (i) during reprecipitation caused by drying after dopant addition and (ii) during the support sintering. The process is evidenced by TEM observation, showing on all the doped samples the presence of palladium crystals encapsulated by a phase that, being still present after careful washing with water, has to be ascribed to the modified support. The fraction of palladium crystals covered by the carrier increases with K/Pd atomic ratio and calcination temperature, at least from a semiquantitative point of view. Figure 8 shows representative images in the case of Pdundand PdK20 before and after washing. The contribution of other processes, if operating, is surely less effective. Pd sintering is negligible in the operating conditions, as demonstrated by TEM,36,37and the direct covering by carbonate salt can be rejected because the encapsulating phase is insoluble
in water, and it tends to increase with temperature while the carbonate decomposes.
The metal encapsulation observed by TEM is quite similar to that observed in the case of strong metal-support interaction (SMSI) effect that is typically associated with reducible supports (i.e., TiO2, CeO2).53,54In that case, the chemisorption capability of the metal particles by probe molecules (i.e., H2or CO) heavily decreases by H2-reduction at high temperature (g773 K),53 due to a particle decoration by the mobilized reduced support. However, care must be done in order to distinguish true SMSI-like phenomena, as originally defined by Tauster et al.,53,54from effects similar to SMSI, but having a different origin.55This is the case of the present system, where the Pd particles coverage is caused by support mobility in absence of H2: a first Pd particle coverage occurs because of reprecipitation of the modified support caused by dopant addition, and a second one occurs because of support sintering.
Thermal treatments in H2(smaller symbols in Figure 7) cause a diminution of dispersion higher than those observed in air, As being equal. This means that other processes not directly related to support changes are operating, such as sintering and poison-ing, as evidenced by TEM (Figure 9a,b) and IR spectroscopy of adsorbed CO (Figure 9c), respectively.36,37
4. Conclusions
The changes in the support morphology, at nano- and microscale level, along the processes involved in the prepara-tion of a supported Pd catalyst have been investigated by adsorption of N2at 77 K and SEM. In particular, the effect of Pd deposition, doping and thermal treatments have been studied. The investigation of Pd properties and of the catalytic
Figure 8. High-resolution TEM images of the particles on Pd/SA catalysts: (a) undoped sample (Pdund); (b) sample PdK20 treated at 823 K in
air; (c) the same sample shown in panel b after washing with water.
Figure 9. TEM images of the particles of Pdund(a) and PdK20 (b) catalysts H2-reduced at 823 K. Part c reports the FTIR spectra of CO
(PCO= 50 Torr) adsorbed on the Pdundsample H2-reduced at increasing temperature.
(53) Tauster, S. J.; Fung, S. C.; Garten, R. L. J. Am. Chem. Soc. 1978, 100, 170– 175.
(54) Tauster, S. J. Acc. Chem. Res. 1987, 20, 389–394.
(55) Bernal, S.; Calvino, J. J.; Cauqui, M. A.; Gatica, J. M.; Lopez Cartes, C.; Perez Omil, J. A.; Pintado, J. M. Catal. Today 2003, 77, 385–406.
performance cannot be performed apart from considering that Pd crystallites are not deposited on an inert and unmodifiable surface.
Among the investigated supports, viz., activated carbons, γ-Al2O3, SiO2, and SiO2-Al2O3(SA), the SA one is extremely plastic and reactive, so it has been considered for additional investigations. We have demonstrated that many processes, summarized in Figure 10, contribute to modify the support morphology at nano- and microscale level.
During the Pd deposition (Figure 10a), the support itself is partially dissolved and removed as a result of both contact with the basic precipitating agent and the final washing. The dissolu-tion acts mainly on irregularities and diaphragms, resulting in a decrease of As and an increase in the pore size, partly true (demolition of diaphragms) and partly apparent (disappearance of the narrow interstices between the irregularities).
When the undoped sample is treated at high temperature (Figure 10b), only modest phenomena are observed. The loss of H2O (in particular between 393 and 673 K) causes the occlusion of some narrow channels connecting contiguous pores, and/or a weak shrinking and smoothing of pores. Consequently, a weak decline of Asand Vpat constant Dmis observed.
During the doping with potassium carbonate (Figure 10c), the support dissolution continues (the greater the carbonate concen-tration, the greater the dissolution extent). In this case, the dissolved material is not removed, but reprecipitates, at least partially, as potassium salt during the subsequent drying at 393 K. This involves the definitive disappearance of irregularities and, probably, also the removal of some pore walls. The reprecipita-tion of support, as well as that of unreacted carbonate, occurs
mainly outside the pores. As a result, there is an enlargement of pores and a decrease of As.
When doped samples are treated at high temperature (Figure 10d,e), the reaction between carbonate and support (with release of CO2) causes the mobilization of the support itself (“alkaline fusion”), with sintering phenomena that can reach the total col-lapse of the porous structure at a temperature that becomes lower as the concentration of potassium carbonate becomes greater.
The investigated support is a good (and quite complete) model of the processes that can occur on material used as a support for noble metal catalysts. One or more of the observed processes, namely, dissolution and removal or reprecipitation of support during catalyst preparation and doping, change of composition, transfer of material from the inner pores to outer particles, change of local morphology with smoothing or destruction of pore walls, shrinking, sintering of support elementary particles, can be observed on most of the supports.
The modification of the support influences, directly or indir-ectly, Pd crystallites, that can be doped or even covered by materials from support, made more or less accessible or even inaccessible by pore narrowing, widening or blocking, as demon-strated by TEM, CO chemisorption, and FTIR spectroscopy of adsorbed CO. As a result, most properties of Pd crystallites, such as sensitivity to sintering, chemical composition, and availability, depend on support modification during catalyst manufacture.
Acknowledgment. We thank Massimo Graziani (Chimet S.p. A) for the N2adsorption and CO chemisorption measurements. The stimulating discussion with L. Bertinetti and A. Zecchina is gratefully acknowledged.
Figure 10. Schematic representation of the entire process starting from deposition of palladium onto SA support to further doping and thermal treatments. The light green represents the support interacted with potassium carbonate; conversely, the light blue indicates the material that remains unchanged.
